Part 27: Euler’s Method for UPSC ISS: The Tangent March and Its Error Bounds

Welcome back to Part 27 of Module 7, updated for the UPSC ISS 2026 to 2027 cycle. Picard’s method is elegant, but it dies the moment the integration turns impossible. When analytical routes close, we march numerically, step by step, and the oldest marching method of all is Euler’s Method. It is delightfully simple to apply, which is exactly why UPSC tests it through its error bounds instead.

The recurrence is $y_{n+1} = y_n + h\,f(x_n, y_n)$: next value equals current value plus step size times current slope. The local error of one step is of order $O(h^2)$, bounded by $\tfrac{h^2}{2}\max|y”|$, while the global error over many steps is $O(h)$.

Exam context: 80 questions, 200 marks, 2 hours, 2.5 marks per question, about 0.83 deducted per wrong answer. The 2025 cycle carried a beautiful Euler question combining the formula with basic calculus, and we solve it fully below.

This post is Part 27 of the Differential Equations Guide (Module 7), inside our UPSC ISS Numerical Analysis Complete Guide. Roadmap at the UPSC ISS hub.

The Story of Rohan and the Chemical Leak

Rohan, an ISS officer managing a hazardous waste facility, faces a leaking chemical tank. The leak rate depends on both the time and the amount remaining, and the equation is too tangled to integrate.

His workaround is geometric. He computes the slope at $x = 0$, walks a short straight line along that tangent to reach $x = 0.1$, computes the new slope there, walks another short line to $x = 0.2$, and repeats until $x = 1$. Approximating a curved solution by a sequence of short straight lines is exactly Euler’s Method.

The Core Concept: The Tangent March

For $\dfrac{dy}{dx} = f(x, y)$ with $y(x_0) = y_0$​, march forward in steps of width $h$:$$y_{n+1} = y_n + h\,f(x_n, y_n)$$

In simple words: next value equals current value plus step size times current slope.

The built in weakness. Straight lines cannot follow curvature, so the marched path drifts steadily away from the true curve. A tiny $h$ restores accuracy but makes the march painfully slow; a large $h$ is fast but wildly inaccurate. This trade off is itself a theory question.

The Ultimate Exam Trap: Local Versus Global Error

Local truncation error, the error of one single step, comes from ignoring the curvature within that step:$$\text{Local error bound} = \frac{h^2}{2}\max\left|y”(x)\right|$$

The order is $O(h^2)$.

Global truncation error, the accumulation across many steps, loses one power of $h$:$$\text{Global error} = O(h)$$

StatChakravyuh Pro Tips

1. The Taylor series bridge. Write $y_{n+1} = y_n + h\,y_n’ + \tfrac{h^2}{2!}y_n” + \dots$ Euler’s method keeps exactly the first two terms and drops the rest, so it agrees with the Taylor series strictly up to the term in $h$.
2. The Runge Kutta identity, heavily repeated. Asked which Runge Kutta method Euler’s method equals, the answer is the Runge Kutta method of the first order.
3. The bounded derivative hack. When the error bound formula $\tfrac{h^2}{2}\max|y”|$ meets a function whose second derivative is bounded by a simple constant, substitute the bound directly instead of computing anything. The bound question is a formula recall question in disguise.

Solved PYQ Masterclass

PYQ (UPSC ISS 2025 cycle, as recorded in our PYQ archive): The forward Euler method finds the numerical solution $y_1$ at $x = 0.2$ with $h = 0.2$ for the initial value problem $\dfrac{dy}{dx} = -2xy^2$, $y(0) = 1$. What is $\left|y(0.2) – y_1\right|$ approximately?

(a) 0.0215 (b) 0.0385 (c) 0.9615 (d) 1

Step 1: The goal. $\left|y(0.2) – y_1\right|$ is the error, exact value minus Euler value.

Step 2: The Euler value. $y_1 = y_0 + h\,f(x_0, y_0) = 1 + 0.2 \times (-2 \cdot 0 \cdot 1^2) = 1 + 0 = 1.0$. The starting slope is zero, so the tangent march stays flat.

Step 3: The exact value. Separate variables: $\dfrac{dy}{y^2} = -2x\,dx$. Integrate: $-\dfrac{1}{y} = -x^2 + C$. Apply $y(0) = 1$: $C = -1$. So $y = \dfrac{1}{x^2 + 1}$, and $y(0.2) = \dfrac{1}{1.04} \approx 0.9615$.

Step 4: The error. $\left|0.9615 – 1.0\right| = 0.0385$.

Final Answer: (b) 0.0385, verified by hand. One question testing the Euler formula, separation of variables, and error literacy simultaneously.

Bonus quick march: for $\dfrac{dy}{dx} = x + y$ with $y(0) = 1$ and $h = 0.1$, the first step gives $y_1 = 1 + 0.1(0 + 1) = 1.1$, and the second gives $y_2 = 1.1 + 0.1(0.1 + 1.1) = 1.22$.

Common Traps to Avoid

Trap 1: Quoting $O(h^2)$ for the global error. Local is $h^2$; global is $h$.

Trap 2: Evaluating the slope at the new point instead of the current one. The forward Euler slope always comes from $(x_n, y_n)$.

Trap 3: Missing that a zero starting slope freezes the first step, exactly the twist of the 2025 cycle question.

Frequently Asked Questions

Take the Next Step with StatChakravyuh

Join the Free UPSC ISS WhatsApp Community. All parts of this series plus the Module 7 Practice PDF with 60 MCQs arrive there first. 👉 Join the WhatsApp Community now

Enroll in the Free UPSC ISS Course. Structured lessons and practice sets on the Practice Improve Repeat system, completely free. 👉 Enroll in the Free Course

Coming up in Part 28: we repair the drifting straight lines with a feedback loop. The Modified Euler’s Method and the Predictor Corrector idea.